JPH03502631A - Transmitting device and receiving device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] Symbols advanced from interframe DPCM for time division multiplexed spatial frequency analysis of video signals Generation of code This invention is a digital image televise. Regarding.

Background of the invention Differential pulse code modulation (DPCM) of digitized video signals is This is often an important step in the digital transmission of the frame order of Gillan images. DP In CM, each signal that describes the video element (pixel) of the rask scan of successive frames is used. The sample is differentially combined with its associated pixel, and the resulting difference is the pulse code. coded using a coding modulation procedure. pixel is the same as the previous sample or one scan line Other samples that are differentially combined or spatially adjacent to the PCM procedures are well known, such that the interframe D It can be classified as a PCM procedure. However, if the pixel is at the corresponding location in the previous frame A DPCM of the form "r inter-frame" that is differentially combined with pixels located in the intra-frame D There are particular advantages compared to PCM procedures. When performing run length coding, zero A long run of DPCM samples with values occurs in a static portion of a series of frames. there is a possibility. Quantization noise is spatial as in intraframe DPCM. This is also temporal in interframe DPCM. Time integral in the human visual system By their nature, these visual systems are less likely to distinguish between temporal quantization noise. centre Error propagation in inter-frame D-PCM coding affects a single pixel. There is a tendency to

In intraframe DPCM, errors propagate over a wide portion of one frame. and can be very noticeable as far as the human visual system is concerned. Video signal frame Before inter-frame DPCM transmission, the video signal is selected into its respective components. In some cases, a procedure to apply a filter effect may be provided.

Generally speaking, the structure of the interframe digital pulse code modulator is as follows. That's right. Digitize the current video signal or video signal components to create a digital form a stream of input samples and feed it to the minuend input port of the subtractor. supply. The subtraction input port of the subtractor receives the preset value of the digital input sample. , the difference output port outputs a digital error signal in response to the samples received at the input port. generate a number. This digital error signal is fed to a quantizer, which converts it into a digital sort the error signal into range bins. The modulator uses the sample value of each bin in the quantizer In many cases, the range is adjustable during operation. quantizer is simply wrong Limiting the uniform range bins by suppressing the lower bits in the signal Alternatively, it may be a format that limits range bins of different sizes. amount The output signal of the child generator is a stream of range bin numbers arranged symmetrically around zero. , which is the output signal of the modulator. Preparation for next frame sample The measured value is obtained as follows. Each sample of the modulator output signal consists of a series of number of bins. sample to the nominal error signal sample value for that number of bins and convert the current digit The digital input that occurs one frame later, added to the predicted value of the digital input sample. The expected value of the sample is reached. Write this predicted value to the frame storage memory and is read to the subtraction input port of the subtracter after programming.

The general structure of an interframe digital pulse code demodulator is as follows: It is. of the demodulator input signal corresponding to the samples of the modulator output signal mentioned in the previous stage. Each sample is a bin number, which is converted to a nominal error signal value for that bin number. and is applied to the first addend power port of the adder. The second addend port of the adder is The predicted value is received and the sum output port of the adder receives the digital pulse code modulation. A digital signal similar to the digital input signal provided to the device is generated. this The digital signal is written to the frame store memory and placed in the second addend port of the adder. After one frame, it is read out as a presoaking 1 value.

"Compressed quantized image data suitable for use in telephone conferences" granted on May 5, 1987. N, J with the name "Data transmission method".

Fedele, A., Acampora, P. J., Bird and R.

Hingorami U.S. Pat. No. 4,663.660 describes an interframe DPCM feedback loop. Spatial frequency analysis of television images performed in octave units within the range of Are listed. The results of this spatial frequency analysis are individually quantized and used in each circuit. The symbolic code converts the digitized video sample into a symbolic code. A digital line with a narrower bandwidth than the digital line required for transmission. time division multiplexed (TDM) for transmission over the network. US patent of Feder et al. The symbol coding scheme described in A statistical formula code such as Huffman coding for the run length of DPCM samples with a value of Use grading.

“2 Minutes” filed on April 17, 1987 and assigned to RCA Corporation. U.S. patent application filed by Nori, N. Schiff entitled ``Resolution Level DPCM System'' No. 040.470, following spatial frequency analysis of television images, The use of different differential pulse code modulators for different analysis results is described. ing. The DPCM signals are individually converted into symbolic codes, and the symbolic codes are It is TDM (time division multiplexed) for transmission over a digital line.

The problem with the system described above is that the speed of the multiplexed symbol codes is not uniform. , the time-division multiplexing of symbolic codes becomes complex. Furthermore, this system The structure becomes significantly more complex, and both require, for example, multiple symbol encoders. At the transmitting station of the digital television transmission equipment that we implemented, uniform code was used. spatial frequency analysis of video signals from television cameras. , or time-division multiplexing of the interframe DPCM response to the interframe DPCM response. Completed before symbol encoding of samples. This allows for variable-length, irregular The problem of time division multiplexing of symbolic codes with fast coding speed is avoided.

Time division multiplexing is a symbol coding process that includes run length encoding of DPCM samples. This is done to facilitate the protocol. Reception of digital television transmission equipment In the signal station, time division decomposition is performed before symbol decoding and is performed line by line.

Brief description of the drawing Figure 1 shows a digital television transmission device that can use this invention. FIG.

FIG. 2 is a block diagram of the transmitting station of the device shown in FIG. This is an implementation of one aspect of the above.

FIG. 3 is a block diagram of a velocity buffer that can act as a line compressor or line expander. Ru.

Figure 4 shows how time-division multiplexing of video signal subbands is performed using the Korg shown in Figure 2. FIG.

FIG. 5 is a block diagram of the Coorg for the transmitting station of the device shown in FIG. It replaces the Korg shown in Figure 2 and embodies one aspect of the invention.

6 and 7 are block diagrams of the decoder of the receiving station of the apparatus shown in FIG. A decoder embodying an aspect of the invention.

FIG. 8 is a block diagram of another Korg embodying one aspect of the invention.

FIG. 9 is a block diagram showing a portion of the Korg in FIG. 8 in detail.

In the block diagram of the drawing, the control signal connections follow the convention of being represented by dashed lines.

detailed explanation In FIG. 1, visible light radiation from the field of view is transmitted to a transmitting station 1 consisting of elements 2 to 8. It is sensed by the revision fist camera 2. The television camera is in color It is shown as being a revision camera, which modulates the structure of the field of view with color selectivity. Generates at least three video signals upon filtering and subsequent light sensing do. These color signals are stored in a color matrix within the television camera 2. A relatively wideband luminance signal Y and a relatively narrowband luminance signal Y are electronically combined in a chrominance signals C1 and C2 are generated. C1 and C2 are the color differences from Y It is ok to describe it. Or C1 and C2 are additive, like red and blue. It may describe narrowband luminance values of two primary colors. rask scanned Y , c+ and C2 analog signals are connected to connection 3. ．． Supplied to Korg 7 via 4.5 Ru. The Korg 7 receives the synchronization signal or timing signal from the television camera 2 via the connection 6. It also receives timing signals.

Korg 7 samples the Y, C+ and C2 signals and digitizes the samples. Contains analog-to-digital conversion circuit. Korg 7 uses digital compression technology to , encodes the digitized Y*C1 and C2 signals. By Korg 7 The digital code generated is supplied to the code transmitter 8, which transmits the The digital code is transmitted via the transmission medium 9. For example, the code transmitter 8 Phase shift keying radio frequency for transmission via satellite link used as communication medium 9 It may be a transmitter.

In FIG. The case of transmitting to receiving station 1G is shown. At the receiving station 10, the code receiver 11 is from the transmission medium 9 using a digital signal similar to that generated by Korg 7. Restore code. The recovered digital code is sent to decoder 12. revenge Unit 12 restores the digital compression brought in by Korg 7 and returns Y, C1 and perform the digital-to-analog conversion necessary to regenerate the C2 analog signal. Now. Color matrix circuit 13 electronically combines the Y, C+ and C2 signals. , generates red (R), blue (B) and green (G) signals. R, G and B signals and decoding The timing signal supplied from the device 12 via the connection 14 is supplied to the color television screen. is applied to the operation monitor 15. This monitor is detected by Television Fist Camera 2 displays an image similar to the field of view on its visible screen.

The television camera 2, the Korg 7, and the code transmitter 8 are connected to stations other than the receiving station 10. The code can be transmitted to the receiving station, for example in broadcast format. FIG. 1 shows the element 11 A second receiving station 20 composed of elements 21 to 25 similar to those of elements 11 to 15; A third receiving station 30 is shown which is composed of elements 31 to 35 similar to those of elements 31 to 15. . Each receiving station 10.20.30 has its own transmitting device similar to elements 2 to 8. The communication station 1 may also have a receiving device similar to the elements 11 to 15. good. This allows for two-way communication between each station, as is commonly desired for video teleconferencing. Direction communication is possible.

Of particular interest as far as this invention is concerned is the form of Korg 7 and the decoder 12 parallelism. and the formats that the corresponding decoders 22 and 32 can take.

FIG. 2 shows one possible form of Korg 7. TV Jijin Camera 2 is Y, The C+ and C2 analog video signals are converted to analog-to-digital converters 41 and 4, respectively. Supply on 2.43. Line scanning speed and transducers 41, 42, 43 per line The number of samples taken in the It is best to choose a value that is twice the standard, and for this purpose a reserve frequency in the spatial frequency domain is Some filtering may be required. This is what the camera 2 lens assembly says. Although this can be done optically to some extent, pre-filtering can be done using an electrical filter. It can be strengthened by filter action. This kind of electrical filtering action is From analog to digital converters 41 to 43 荊1: Performed in the analog domain Should. Electrical filtering is done in a separable manner, with horizontal filters The router action is also easy to perform. In the vertical direction perpendicular to the scan line, the camera The wobbling of the scan line position in step 2 shifts the delay line filtering action to the analog domain. It can be done with

The relatively narrowband C1 and C2 signals are sampled at a lower sampling rate than the relatively wideband Y signal. It is preferable to sample this low sampling rate at a low multiple of the Y sampling rate. It is convenient to choose. 2 of the sampling rate of Y, where n is a positive integer in both cases. It is particularly convenient to use subsampling rates that are low multiples. Integer n is 1, 2 ．． You can choose 3 or 4. C1 and C2 are like European TV broadcasts Even if R-Y and B-Y, %CI and C2 are and Q, or even if C1 and C2 are narrowband R and B, set n to 2. Then, if the Y sampling density is sufficient, sufficiently dense sampling of the C1 and C2 signals will be obtained. come. In order to set n to 3, from the viewpoint of image quality, the sampling period of Y must be increased for each frame. It is preferable to alternate the spatial phases of the sampling period of C1 and C2 for each period. Indeed, this phase alternation allows samples that differ in time by one frame to be spatially separated. The calculation of DPCM samples requires a spatial interpolation procedure to match the becomes somewhat complicated. If n is 4, the chrominance interval for each frame is Phase alternation for racing is essential for satisfactory image quality.

Y at least twice the Nyquist φ ray in either direction.

sampling each of the C1 and C2 analog video signals (this (known to the inventor from Dr. E. H. Adelson) is a method that Signal amplitude can be quantized using digital processing without noticeable reasing effects. . For a static image, more pixel samples are used for a given amount of spatial resolution. However, the run of DPCM samples with zero value based on these pixel samples is one The layers may become longer. Number of pixel samples when using run length coding increases, so that non-zero DPCM samples as well as zero-valued DPCM Statistical formula code based on sample runs is much longer.

The broadband digitized Y signal from the analog-to-digital converter 41 is is applied to the input port of a digital delay line 44, which provides a narrowband signal throughout the device. Compensates for the delay that occurs when processing the C1 and C2 signals. Digital delay line is an example For example, it consists of random access memory with staggered read and write addresses. I can do it. The delay of the C1 and C2 signals, especially in the direction perpendicular to the scan line, This occurs when these signals are spatially frequency filtered to a narrower bandwidth. Further delays occur in the time division multiplexing of the spatial frequency analysis of the C1 and C2 signals. This many The weighting will be explained later.

digital delay line 44, analog-to-digital converter 42 and analog-to-digital converter 42; The respective digitized Y, CI and C2 signals from the digital converter 43 are It is applied as an input signal to the spatial frequency analyzers 45, 46, and 47. octave A so-called “part pyramid” pie that performs two-dimensional spatial frequency analysis using bandwidth. Pline type spatial frequency analyzer is called "real-time hierarchical pyramid type signal processing device". was transferred to RCA Corporation on June 16, 1987 under the name C.R. , Carlson, J.H., Alllebaita, and R.H., Bessler, USA. It is known from patent no. 4,674°125. The so-called "Anderson FSD" The pyramid J-shaped spatial frequency analysis was filed on May 15, 1987, and RC Pending U.S. Patent Application for C, H, Underrun Assigned to A Corporation It is listed in serial number 051.210.

L.N. Schiff, in U.S. Patent Application Serial No. 040,470, in the horizontal direction, i.e. in most applications, simply the upper and lower spatial frequencies. A spatial analyzer that separates several spectra is described. Which form of space mentioned above? Although frequency analyzers can also be used in various embodiments of the invention, spatial frequency analysis 45, 46, 47 are right angle mirrors to separate the input signal into four subbands. ・It is shown as a type that uses a filter. These subbands are each empty The number is reduced to 2=1 in the inter-direction. Readers will be able to read the October 1986 IEEE Tra. Acoustics on Acoustics, Speech-and-Signal Pro Locessing, appearing on pages 1278 to 2388 of ASSP-34. Paper by J. W. Woods and S. D. O'Neill, "Subband Coding of Images" Please refer to

Spatial position of edges and details when sampled at least twice the Nyquist Ray. The sensitivity of the filter's response to phase decreases, thus reducing the sensitivity of the filter to moving edges and details. Sources of aliasing are eliminated. High spatial frequency filter in spatial frequency analysis procedure When heterodyning the response to baseband, the error in the heterodyning procedure In order to avoid ashing, this filtering action is performed as a band spatial frequency filtering action. It is preferable that The band-pass spatial filtering effect is due to the aliasing effect. introduced by quantization effects associated with digitization of the signal, which can lead to undesirable results. Suppresses the response above the band. of the camera video signal before spatial frequency analysis. Pre-filtering can also be used, especially if the high-pass filter response is heterogeneous to baseband. When rodining, to prevent responses above the band that would cause aliasing. It is done. The right-angle mirror filtering followed by a reduction step down to baseband. It relies on terodyne action.

A spatial frequency analyzer 45 uses a first pair of right angle mirror filters to determine the spatial frequency. The wavenumber filter effect converts the Y signal into upper and lower half bands in the first direction. Then, the spectrum of half of each band is directly divided into two bands. A filter effect is applied to the upper and lower spectra as components in the second direction that intersect with each other. and separate. This filtering action to obtain the 1/4 band spectrum is Assume that it is done in brine form. One of the first and second directions is a rask run. is parallel to the scanning line of the scanned Y signal, and the other of the first and second directions is opposite to the scanning line. and horizontal direction.

Using a pair of right-angle mirror filters with odd symmetry in each dimension, , the component spectra are preferably reduced 2:1 with spatial antiphase. child Alternatively, an equivalent non-separable right-angle mirror filter action may be used. stomach. On every other scan line of the Y input signal, the spatial frequency analyzer 45 calculates the interframe DP. The signal "LL" is supplied to the CM modulator 48. The signal "LL" is applied both vertically and horizontally. However, it is the response of a spatial frequency filter that has a low-pass property, and the The number of samples is half of Y. Between the same every other scan line of the Y signal, the spatial frequency solution Analyzer 45 supplies signal Y to interframe DPCM modulator 49. The signal Y□L is , L A spatial frequency filter that is vertically low-pass but horizontally high-pass. response of the router, where the number of samples per scan line is the same as Y, and for each pixel " It is temporally interleaved with LL. The frequency synthesizer 45 is Y,! 11 times and Y Samples of the HH signal are sent to the DPCM modulators 50 and 51 respectively during alternate Y signal scan lines. supply to. These alternating signal scanning lines are used by the frequency synthesizer 45 as Y and YH. Sample L of L signal It is interleaved with the scanning line of the Y signal that supplies the signal. Y, H and "H)] signal are the responses of spatial frequency filters that are both vertically high-pass in nature.

YLH has low-pass characteristics in the horizontal direction, and YHH has high-pass characteristics in the horizontal direction. have. For both Y and HH, the number of samples per scanning line is L)I It is half of the Y signal and is temporally interleaved with each other for each pixel. YY, Y And the frame per frame of YHH signal HL 1. )I The total number of samples per frame is equal to the number of samples per frame of the Y signal. .

Y, Y, Y is 3/4 of Y per frame I11, 1, H )I)I Having only the number of pulls means that the two-dimensional right-angle mirror filter action reduces Y to one frame. LL low frequency component with 174 samples of Y per frame and Y per frame. Another reduction method that separates into complementary high-frequency components with the same number of samples as This is one reason why it is preferable to multi-step subband coding schemes. . Furthermore, Y, Y, Y Shin HL LH) IH The zero value tends to have a longer run than the high-frequency components of Y. There is a direction.

Four sub-spectrums “LL’YY, Y” by the spatial frequency analyzer 45 Spatial frequency analysis calculates the spatial frequency) IL' L) I) 1)! Analogous to the spatial frequency analysis of CI and C2 performed by analyzers 46.47 There is. The CI is analyzed by a spatial frequency analyzer 46 to generate a C signal, which is horizontal Also LL in the direction It also has a low-pass property in the vertical direction, and can be used as an input signal to the DPCM modulator 52. will be supplied. C signal is horizontal 1) IL It is a high pass in the direction and a low pass in the vertical direction, but for the DPCM modulator 53. is supplied as an input signal.

C signal is low pass in the horizontal direction and IL) I in the vertical direction. is high-pass, but is provided as an input signal to the DPCM modulator 54. . The C signal is applied both horizontally and vertically to the vertical blade IH1 ( It is also high-pass in the direction and is supplied as an input signal to the DPCM modulator 55. Ru. C2 is analyzed by a spatial frequency analyzer 46 and output to DPCM modulators 56 and 57, respectively. ．． C, C supplied as input signal to 58.59.

2LL 2HL Generates C and C signals. 2L)I 2H) from analyzer 46.47] The result of the spatial frequency analysis is lower than the result of the spatial frequency analysis from the analyzer 45. However, the rate of decrease is determined by the rate of C1 and C2 relative to the sampling rate of Y. Same as sampling speed.

The DPCM signals generated by the modulators 48 to 59 are sent to the time division multiplexer 6. 0 inputs and this multiplexer interleaves them line by line. the continuous samples as an input signal to the run length code assembler 62. - Supply a stream. Assembler 62 detects a sample with a value of zero , the run length of these samples is counted. The number of samples with such a value of zero is If the length of the input signal exceeds a non-zero value in the input signal supplied to the assembler 62, converted to a range and entered into the statistical formula coding Korg lookup table 63. is supplied as an address to a zero-value sample in the sample stream. Used in place of the pull itself. Most of LUT 63 is fixed memory (ROM) etc. This can be realized by a code book storage device located in

As an example, the code in Korg Table 63 is Huffman code, more frequently occurring run lengths of zero values, and common zero values. A prefix indicating that run lengths that are not specified should be specified in direct binary coding form. This may be a code switching flag used to generate a code. Prefix code and The binary coding of an unusual run length of zero value is R in Korg Table 63. Keep the dimensions of the OM within reasonable limits. The codes in Table 63 are of various bit lengths. and is fed to a parallel-to-serial converter 64, which converts the code The bits are applied serially to the code output speed buffer 65. Speed buffer 65 is first It is a first-in, first-out (FIFO) format.

These bits are serially supplied to the code output speed buffer 65 at irregular intervals. However, this is only because the output codes of table 63 have various bit lengths. This is also due to timing irregularities caused by the run length encoding process.

Although these periods occur irregularly, YLL'Y)IL'Y L) I’　YHl(”ILI,　　　　　C, C, C1)IL　　　IL HIIIH’ Time division multiplexing of C, C, C, C is tied 2LL 2) IL 2LI+ 211) Before processing that introduces 1-timing irregularities, the multiplexer 60 Because this is done, the order of the serial bits is always correct. timing irregularities What this results in is that the code stream fed to the code output rate buffer 65 This is because a gap is created between the parts. The code output speed buffer 65 is Adjust the timing of the bits to eliminate gaps, or at least many of them. , digital transmission by a code transmitter receiving the code output from the speed buffer 65. This further compresses the transmission time.

The output code velocity buffer 65 is configured to , many bits coming in during the course of a frame with a large change from the previous frame. It must have sufficient storage capacity to accommodate the number of cards. Shown in Figure 2 In similar systems, elements 62-65 are not in the DPCM loop, but are in the velocity buffer. The memory capacity of the 65 is considerably larger than the average frame, making it economically viable. For example, in the worst case, it can be reduced to a few frames. In this case, the abnormally high An adaptive code that changes the coding procedure when a high bit rate persists for a period of time. It is possible to avoid using the editing method. When choosing a design like this, , there is no fill control 66 for the output code rate buffer 65. This method is fast large code output to avoid having to provide filler controls for the buffer. However, it may be satisfactory for a one-way video transmission scheme. Naturally However, in two-way video transmission systems, i.e., teleconferencing, a large cord output at the transmitting station is used. Delays due to the speed buffer and the complementary code input speed buffer at the receiving station are allowed. This delay can lead to other substantial transmission delays, such as satellite links. This is especially true if it increases the delay due to transmission over the network. This is for video display By appropriately delaying the audio transmission that accompanies it, it becomes less of a problem. It is possible to make it disappear. However, the delay of 1710 seconds or more is more important than the ability of the speaker to participate in normal conversational exchanges. This is problematic because you have to start talking on the queue.

In general, the bit rate for the speed buffer 65 is reduced so that large motion The code output is performed using a controller 66 that changes the coding procedure during the frame. It is desirable to control the filling of force velocity buffer 65. By doing this, the speed buffer As the storage requirements for the complementary speed buffers in the buffer 65 and the receiving station become more relaxed, , the delay caused by these speed buffers is also shorter. The speed buffer 65 Collect the serial bit inputs into words of uniform bit length and Each successive word is stored in random access memory (RAM) inside the module. A formula that stores in the word storage location with the next highest write address. It can be formatted. At the same time, successive read addresses of word locations in RAM are Reads RAM at regular word rate with irregular scanning A parallel-to-serial converter inside the speed buffer 65 converts the code into a serial type. is converted back into the equation, thus producing a uniform speed serial output code. speed bag Unless the file 65 is overfilled (or underfilled), the read address is It lags behind the write address, but the amount changes, and the RAM write and read address The difference between is an indication of the amount of storage capacity remaining in the RAM. This display is fast The data is supplied by a buffer 65 via a connection 67 to a filling controller 66 . filling The action of controller 66 is such that velocity buffer 65 remains fairly close to empty most of the time. However, you may have to sacrifice frames with a lot of movement, or at least that's usually the case. The first thing to do is to adjust the coding method to avoid this. buried controller There are many ways in which 66 can perform its function.

For example, if there is a problem with the code output speed buffer 65, which has a tendency to overfill. If so, the padding controller 66 sends control signals to the interframe DPCM modulators 48 to 59. Therefore, it is possible to configure the elements of each quantizer to be roughly divided into bins. Special In addition, the bins with zero values are made wider to accommodate the scan lines of DPCM modulators 48-59. Lengthens the run of zero-valued samples, which may occur more often in the output. DPC The change in the bin width of the quantizer elements in the M modulators 48 to 59 is controlled by a filling controller 66. sends a signal to Korg's LUT 63 via connection 68 and outputs the modified DPCM signal. achieved by adjusting the coding table to better fit the statistics of can be achieved. Adjustment of the coding table requires different codes from ROM. The easiest way to do this is by choosing a group table.

In bidirectional digital video transmission systems, looka- ries for interframe DPCM If you wish to avoid frame delays associated with the head method, interframe DPC By coarsening the binning in the M modulators 48 to 59, the speed buffer 6 It is possible to quickly respond to overfilling of 5. In-frame corrections are made at the receiving station. (e.g. between blocks of image information) so as not to disturb the operation of the spatial frequency interpolator. It can be done quickly. The transmitting station supplies the code from Cog to the code transmitter in Figure 2. Insert a code word into the data thridome that indicates the nature of the binning change. You can enter. This insertion is done at appropriate intervals between blocks of image information. , so that the decoder at the receiver responds to changes in the decoder's lookup table. You can.

Another option that can be used to avoid overflowing the code output speed buffer 65 The method - in areas where motion occurs in successive image frames, a higher spatial frequency It is a well-known fact that the lower the spatial frequency, the less resolved spatially or temporally. Use the fruit. It should not be zero, which would cause overfilling of the code output speed buffer 65. During frames with many high DPCM values, the spatial frequency of the Y video signal is higher. component, and possibly also the C1 and C2 video signals, in such a way as to suppress their transmission. Configure. The mechanism that does this is not specifically shown in Figure 2, but it is as follows. This is because the procedure mentioned above is incidental to the scope of the claims. However, the behavior of this mechanism The work is as follows.

When an overflow condition in the output rate buffer 65 is predicted, DPCM modulator 49, 50.51 (7) Y, Y, Y (and possibly frame Between rooms D) IL LHHH PCM modulator 53.54.55.57.58.59 C, C, C, C, C, C) The forecast 1) IL　　　ILHIHII　　　2HL　　　2LH2HH measurement value Set it to zero legally. The output signal from the affected DPCM modulator is also forced to zero Make it. This is done by a suitable multiplexing scheme in the DPCM modulator. child A digital umbrella code word is inserted between frames to indicate that an action is occurring. (or the period between blocks of image data), the Korg lookup table It is inserted into the input signal to the transmitter 63 to indicate that such an operation is being performed at the transmitting station. and inform the receiving station. At that time, the receiving station uses its DPCM demodulation circuit to , Y, and in some cases CIHL') fL Lll) I) Strong control of predicted values of Ic, c, c, c, c H1) 1) 1 2) IL 2LH2HH zero, Y, CI and C2 Feeds the Y, CI and C2 spatial frequency synthesizer used to generate the video signal. Replace appropriate locations of the signal with zeros. The zero stuffing procedure explained in this article (this Described in pending U.S. Patent Application Serial No. 040.470 by Hari, N. Schiff Note that this reduces the spatial resolution of the image. . However, this zero-filling procedure removes higher spatial frequency information in the predicted image. and avoid aliasing, which is highly visible to human visual systems. This alias ng appears as a ``ghost edge.'' Stuffing with zeros is extremely harmful to the human visual system. Aliasing does not occur in areas of space that are easily visible. Instead, the bandwidth There is a temporary decrease in , but it is less noticeable and corrects over time.

FIG. 3 shows a linear compression speed buffer in time division multiplexer 60, e.g. The line compression in Figure 3 shows how each of the files 60-1 can be configured. Successive scan lines of spatial frequency analysis provided as input signal to velocity buffer starts from 1, using the number of scan lines per frame as the base number. one after another Assume that an ordinal number can be assigned by . Linear velocity write input multiplexer 70 The odd numbered scanning lines of the frequency analysis are input/output ports of the random access memory 71. line and write to it, and even-numbered scan lines are randomly accessed. It is directed to the human power/output bus line of the bus memory 72 and written there. RAM71 and 72 both have a storage capacity for samples of one scanning line. Noble letter with writing If there is a signal pulse, it is selected by the linear velocity write enable multiplexer 73. It is sent to one of the RAMs 71 and 72 written in the scanning line, and the read code is If there is a signal pulse, it is transferred by the line readability multiplexer 74 to its scanning During the line, it is sent to the other RAM 71.72. Linear velocity reading output multiplexer 7 5 is RAM 71.72 selected for reading and not selected for writing. Connect one input/output bus of the Ru. Linear velocity address multiplexer 76 controls write address generator 77 to Connect to one of the enabled RAMs 71 and 72 and read address generator 78. Connect to one of read-enabled RAMs 71 and 72. From write address Both the generator 77 and the read address generator 78 scan the address generator. child The address generator generates successive addresses of pixel samples along the scan line. However, the speed at which successive addresses are scanned is different for the two address generators.

The write address generator 77 is time division multiplexed as is done in the Korg of FIG. If this allows, it should be shared with other linear velocity buffers operating at the same sampling rate. I can do it. Each write enable signal received by the velocity buffer from fill controller 66 Controls time division multiplexing as far as reading from that RAM is concerned. Figure 3 Hardware The wear acts as a linear velocity buffer. This is because the read address is read intermittently. The speed supplied to take address generator 78 is such that the write address is Because it is faster than the speed that is intermittently supplied from the generator 77, as well as writing and reading capabilities. This is due to the commensurate difference in the pulse speed of the output. One scan line of spatial frequency analysis The sample written to one of RAMs 71 and 72 during this period is in the second half of the scanning line. Read out. The hardware in Figure 3 instead has the read address When the write address is set to the speed that is intermittently supplied by the speed generator 78, line expansion by slowing the rate that is intermittently supplied by the speed generator 77. It can also be used to act as a speed buffer.

FIG. 4 is a time diagram showing how the time division multiplexer 60 operates. This will help you understand how the controller 66 in Figure 2 operates. Also useful.

The progress of the scanning line from camera 2 is on the vertical axis, and the part that carries one frame of image 8 lines per cycle of time-division multiplexer operation that is repeated continuously for minutes. There is a line.

The cycle number N is 0 for the first cycle and only 1 for each successive cycle. The number will be increased. A frame consisting of 240 lines is based on the time division map shown in the diagram in Figure 4. There are 30 cycles of multiplexer operation, and N increases from 0 to 29 in successive cycles. counted. The time between one scan line of video from camera 2 is shown on the horizontal axis.

For example, during the (8N+1)th line, after a compressed scan line of Y samples, Y LHS L A compressed scanline of the file follows. C1 and C2 are both horizontal and vertical directions It is shown that both are sampled at a ratio of 4:1 with respect to Y. By this The run length code assembler 62 of the output signal of the time division multiplexer 60 A simple packing form can be obtained for

If the speed buffer starts to get too full, C1C2HHIHH The scan lines of the If the filling speed is low enough, replace selectively. speed buffer continues C, C and 2) 1. L IHL Scan lines in the speed If the buffer fills more continuously, or if the filling rate is very high, C” ILH and LH The YLH scan line can also be replaced with a scan line having a value of zero. run length code - In response to each scan line having a value of zero, the assembler 62 assembles the Korg LUT. 63, statistical formula code with short bit length for converter 64 from parallel to series Provides an address that selects the output signal. The bit length that frequently occurs like this Shorter statistical formulas reduce the overall code length of the code, substantially reducing the code output The filling speed of the speed buffer 65 is reduced.

By the means described in the previous paragraph, the signal has zero values in different spatial frequency subbands. If the sample runs are roughly the same, the filling speed of the speed buffer 65 is almost 1/ It goes down to 4.

QMF spatial frequency analysis, which divides the signal into more than four subbands, By further reducing the filling speed of the filter 65, even finer tuning of filling control can be achieved. Y signal Further subband coding requires that Y is much larger than C1 and C2. This may be of particular interest because of its sample density.

If you omit the Y sample, you will also omit the sparser CI and C2 samples. This also has a more rapid effect on reducing overfilling of the speed buffer 65. It would be recognized that Reduce only the spatial resolution of Y and use subsampling. speed buffer 6 until Y is at least as sampling density as C1 and C2. It is also possible to change the design to control the filling of 5.

Still, all samples except Y, C and C2LL LL+, Even if the code output speed buffer 65 is replaced with a value of zero, the code output speed buffer 65 tends to become overfilled. If there is a problem with low-pass spatial filtering, further subsampling A configuration that further reduces the spatial resolution of these signals by performing pulling. can be provided. This was granted on October 13, 1987. ``Velocity Buffer Control of Predicted Signal Subtraction and Interpolation for Shaped Pulse Code Modulators'' A. This is described in Patent No. 4,700,226. This configuration Instead of or in addition to changing the binning of the quantizer elements of the M modulators 48-59. It can be added and used.

If it is sometimes necessary to further reduce the resolution of the image, there are facilities to facilitate that procedure. An advantageous method is to use another right-angle mirror filter action to reduce the YC and C1/4 smooth L L’ ILL 2LL Each of the vectors is divided into four 1/1s, each reduced to 2=1 in each spatial direction. It is divided into one set of six spectra. This 1/16 spectrum each has a respective interframe DPCM modulator. multiplexer 60 Change the speed LL ILL 2LL for Y, C and C The buffer is further compressed temporally by a 4:1 ratio of the scan lines, and is time-division multiplexed. , we replace them with four speed buffers each with appropriate delays. In that case, There is a high degree of motion in the series of image frames being transmitted, and Y, C and LL I LL If it becomes necessary to further truncate the transmission beyond just the C2LL being transmitted, The 1/16 spectrum sample with high spatial frequency is replaced with a zero scan line and further processed. It is possible to realize compressed transmission. Appropriate zero stuffing means between frames D Used in PCM modulators, older high spatial frequency information is discarded in areas with movement; Then use the ideas mentioned earlier. Compensation for unsplit 1/4 spectrum delay Of course, we also have to do this, but instead we divide the entire 1/4 spectrum. Therefore, all problems of discriminatory time delay can be avoided.

Korg's explanation in Figure 2 omits an important point in transmitting station design. . A means shall be provided to transmit video synchronization information and zero padding instructions to the receiving station. However, normally if some means of transmitting audio simultaneously can be provided. do not have. The simple design procedure is as follows. Korg lookup table 63 The frame start and frame end codes are time-division multiplexed to the input of the frame. (running When using intraframe synchronization (such as a set of line change codes), the intraframe synchronization symbol is input to the Korg lookup table 63 between the compressed scan lines. Can be divided and multiplexed. Each frame of DPCM samples is coded When finished, the output code rate buffer 65 has the time division multiplexer 60 DP Before starting to supply the next frame of CM samples, the bit stream describing the audio is The stream segments are time-division multiplexed so that the bits provided to rate buffer 65 are Configure it so that there is enough space to put it in To Suri Dream. audio (or multiple Do not mix too many intraframe synchronization symbols (intraframe synchronization symbols) within the time slot of a video frame. is preferable. In this way, Y, Y, Y of the input signal of the assembler 62 , Y, C, C, C.

LL HL LHHHILL IHL IL) IC, C, C, Between C and C segments IHH2LL 2HL 2LH2) i) Concatenate the run lengths of zero values of 1 It will become even better.

Each successive frame DPCM encoded by Cog in Figure 2 represents one component of the image. Note that it is not a single video signal of the rask scan that describes . That's right rather, each successive frame constitutes a successive block of packed image data. each successive block covers all of the image over many rask scan lines. (each of its component primaries). The criteria for packing image data is Keeps the number of rask scan lines written per block of image data small That's true. This reduces the speed buffer required for time division multiplexing to pack image data. The magnitude of the effect and differential delay is reduced. Right angle mirror filter action The subsequent reduction operation is particularly suitable for image packing.

The total number of narrowband color samples in C1 and C2 is the same as the broadband color samples in Y. approximately 1/8 of the number of luminance samples and therefore the retrace period of television camera 2 It is possible to fill in the lost scan lines between the lines. Spatial frequency analysis is performed on the original video signal. You will never generate more or fewer samples than you did ( However, the subsequent spatial frequency analysis step increases the delay). Therefore, The embedded image data block is an area of pixels in the original image from which the data was taken. However, it continues to have a general spatial correspondence or mapping relationship.

If you are a person skilled in the art of installing digital equipment, *25! :I's Korg 2 is this launch. It will be understood that certain changes may be made within the scope of the description. Figure 2 shows the Linear pressure linear velocity buffer 60-1 lowers the sampling rate at which regulators 48 to 59 operate. ...indicates the generation of DPCM using modulators 48 to 59 before 60-12. ing. Instead, the spatial frequency analyzer 45, 46, 47 directly analyzes each analysis result. Directly, the input signal to the linear pressure linear velocity buffer 60-1...60-12 each frame-to-frame DPCM operating at the respective faster sample rate. The modulator calculates the run length from each linear pressure linear velocity buffer 60-1, 60-2. be inserted into the output connection to the input signal bus of the code assembler 62 You can also do that.

The time division multiplexing system shown diagrammatically in Fig. The delay is achieved using a linear pressure linear velocity buffer as shown in Figure 3 (with compensation of course). It is attractive in that it does not require a separate delay line (with the exception of delay line 44). time division If you are willing to use more complex differential delaying of multiplexed signals, you can use this invention. Within range, further various alternating time division multiplexing schemes can be used. say this Some alternatives offer certain operational advantages, but at the cost of more complex hardware. Requires air. For example, an image that is time-division multiplexed in units of (compressed) scan lines. Continuing to assume that the block of data represents eight scan lines of camera 2's video signal, 4, the multiplexing scheme in Figure 4 consists of YL, C four compressed scan lines, and C compressed scan lines. ILL compressed scan line and C 2LL compressed scan It can be modified to connect lines.

As we did for the LL spectrum, we also do the same for the LH, HL and HE spectra. Various types of connections can be made. By doing this, the processing record of the FIFO speed buffer will be A controller for the signal code simplifies generating control signals.

Figure 5 shows a simplified version of the structure of the Coorg in Figure 2, with a large number of inter-frame DPCMs. The modulator is replaced by one DPCM modulator 61 operating at a fast sample rate. ing. One DPCM modulator 61 acts in place of the DPCM modulators 48 to 59. Next, consider the reason why this can be achieved satisfactorily.

By spatial frequency analysis (in one spatial dimension) of variables describing brightness, One or more analysis spectra or subbands result, and at the knee the sample It can take positive or negative values and produces residual low-pass spectra or subbands. Similarly, at knee, the sample always has a positive value. Spatial circumference of variables describing color difference Wavenumber analysis (in one spatial dimension) allows one or more subbands to be Bands are generated, and at the knee the sample can take on positive or negative values, as well as residual low frequency A band occurs, and at the knee the samples can take on positive or negative values. in one dimension The band subbands in which the sample polarity alternates between spatially adjacent samples are (if this trend follows the right-angle mirror filter procedure, then the band sub-band The zeros are masked by the 2:1 reduction procedure used in the band heterogain operation. average frequency). Especially when the camera position remains constant and there is a lot of movement in the field of view. When there is no overlapping component, the pixel sun at the same position in the field of view of the television camera There is a significant tendency for inter-frame correlation between pulls. Therefore, the interframe DPC When the M signal is applied to a spatial frequency analysis in one spatial dimension, all these The statistics of the DPCM response for subband signals of different shapes tend to be similar; It is much stronger than the subband signal statistics of Eq. two orthogonal spatial orders Although the analysis is somewhat more complex when performing the original spatial analysis, The interframe DPCM response statistics for each type of subband signal are also similar. It can be shown that there is a tendency to Synthesis of subband Δ-frame DPCM The totals are similar enough that using the same Korg lookup φ table configuration, , all of which can be encoded statistically.

However, the bits sent from the padding controller 66 to the DPCM modulator 61 via connection 69 The control signal is applied to each compressed scan line segment in the input data stream. It can be changed to Sent from the embedded controller 66 to Korg's LUT 63 The Coorg table adjustment signal is also changed in the same way to the quantizer of the DPCM modulator 61. can handle changes in binning.

Zero stuffing is also performed in a time division multiplexed manner in the interframe DPCM modulator 61. Must. from the modulator 61 to the run length code assembler 62. Both the value of the output output and the predicted value stored in the DPCM feedback connection of modulator 61 are When not transmitting scanning lines with higher wavenumber components, it must be forced to zero. stomach.

Figure 6 shows a digital ψ television transmission system using the Cog shown in Figure 5 as a transmitting station. Indicates the decoder used at the receiving station.

The code input bit stream from the code receiver (not shown in the drawing) is parsed. 81. Parser 81 converts the bit stream into variable length code words. , organize them into a standard bit length, and then perform statistical expression coding decoding. It is applied as an address input signal to the address lookup table 82. child The table in Figure 5 shows the statistical formula coding Korg lookup in Korg. It is complementary to the table 63. Lookup table 82 Therefore, the generated modified DP CM sample is first-in-first-out (FIFO) speed. The data is supplied to the buffer write memory 83. The FIFO buffer 83 stores the run length code. Modified DPCM samples, one at a time, in response to commands from the unbalanced force 84 Put it into an unbalanced force. Unbalanced force 84 is modified DP CM between normal frames Convert to DPCM signal.

The interframe DPCM signal from the unbalanced power 84 is sent to the interframe DPCM demodulator 85. This demodulator converts the spatial frequency analysis sample into a time-multiplexed demodulator. is supplied to the multiplexer 90. This demultiplexer has a linear expansion speed buffer 90 -1.90-2.90-3.90-4.90-5.90-6.90-7.90- 8. It is composed of GO-9, 90-10, 90-1, 1 and 90=12. line extension Speed buffers 90-1.90-2.90-3 and 90-4 are YY, Y and Y l(ll signal, spatial frequency analyzer 45LL' HL LH is supplied to a spatial frequency synthesizer 91 that complements the signal and generates the digitized Y signal again. supply. Line expansion speed buffers 90-5.90-6.90-7 and 9G-8 are C, C.

ILL IHL C and C1HH signals are complemented by the spatial frequency analyzer 46 and LH and supplies the digitized C1 signal to a spatial frequency synthesizer 92 that generates it again. . The linear expansion speed buffers 90-9.90-10.90-11 and 90-12 are C, C.

2LL 2HL C and C signals 1. Complementing the spatial frequency analyzer 47, 2LH2HH and supplies the digitized C2 signal to a spatial frequency synthesizer 93 that generates it again. .

Each of the spatial frequency synthesizers 91, 92, 93 has subbands supplied to its input. The component signals are the respective output signals Y.

Multiple spatial frequency complements resampling up to full sampling rate for C1 or C2. I have an intervening organ. This resampling is well known to those skilled in the art; Here's how to do it: The first step is in the expanded spatial matrix of each subband, The subbands are not filled with the corresponding actual samples from the swarza space matrix. Inserts a zero sample with a value of zero at the new sample position. This first step is a spatial frequency carrier modulation process that generates a low-band and band spatial frequency spectrum. be. The second step uses filtering to remove defects caused by the first step. It is to suppress the spectrum of desires. (This filtering effect is due to the spatial circumference of Korg. If the wavenumber analyzer uses right-angle mirror filter action to perform spatial analysis, then This is a right-angle mirror filter effect. ] The third step is the interpolation result from the second step. are added together, and the result has similar sampling density. spatial frequency When seeking the spatial frequency synthesis of a signal subjected to pyramidal reduction during analysis, the above Repeat the steps described in .

A spatial frequency synthesizer 92 resamples the combined C1 signal and generates a combined Y signal. includes another spatial interpolation step to have as many scan lines as symbols. Similarly, A spatial frequency synthesizer 93 resamples the combined C2 signal and generates a combined Y signal. includes another spatial interpolation step to have as many scan lines as symbols. Another thing like this It is convenient to perform spatial interpolation in two dimensions: horizontally and vertically. It is advantageous. Digitize the color matrix effect of the combined Y, CI and C2 signals. This is especially true if you are trying to do it directly. Korg's compensation delay device! 44 is The delay associated with such separate spatial interpolation of the C1 and C2 signals can be taken into account. I can do it. (Detailed explanation of this invention in which Y delay compensation is performed only by Korg) However, in another embodiment of the invention, the delay compensation may instead be performed at the decoder. Alternatively, it may be distributed between the Korg and the decoder. ) Digital to analog converter 94, 95, and 96 address the regenerated and digitized Y, CI, and C2 signals. Converts to analog Y, CI and C2 signals, and these analog signals are converted into color matrices. The signal is supplied to the logic circuit 97. Color matrix circuit 97 is analog Y, C Color red (R), green (G) and blue (B) drive signals in response to I and C2 signals - Supplied to the monitor receiver 98. A color monitor receiver 98 is shown in FIG. It also receives synchronization and audio signals from other parts of the decoder.

(Alternatively, the color matrix operation is carried out by spatial frequency synthesizers 91, 92, 93 The digital Y, CI and C2 signals from It can also be done in the digital domain by generating the and B signals. the Then convert the digital R, G and B signals using a suitable digital to analog converter. converts the signal into analog R, G, and B signals and applies them to the color monitor receiver 98. Rukoto can. Easier to adjust color saturation and requires fast digital multiplication Color matrix operation in the analog domain is preferable in that it does not ) system control 86 at the beginning or end of a certain code, frame, loaded into speed buffer 83. synchronization codes to indicate the start or end of a scan line, and other uses. What kind of synchronization code is used and what coding method is used by Coorg? Decode the code that shows. These cords are provided to the controller 86 via connection 87. be provided. Controller 86 is responsive to the synchronization code to cause controller 86 to perform time division demultiplexing. A series of write functions to be transmitted to RAM in the linear expansion speed buffer of Lexa 9G. signals A, B, C, D, E, F, G, H and a series of readability signals P, Q, R, S, Timing T, U, V, and W. The controller 86 connects the LUT of the decoder via the connection 88 Supplying a control signal to 82 to modify the operating characteristics of Korg's LUT 63 Here is a code showing that a similar operation was performed in Korg: Frame Rate Buffer 8 3, the controller 86 changes the operating characteristics of the LUT 82.

The controller 86 restores the zero stuffing command received from the speed buffer 83, if any. and sends a command to the DPCM demodulator 85 via the connection 89 to A circuit inserts a zero into the appropriate segment of the predicted signal stored in demodulator 85. In addition to packing the predicted value, the output signal of the demodulator 85 is forced to have a value of zero. Make it.

The decoder in Fig. 7 is the same as the decoder in Fig. 6 in order to reduce the sampling rate of the DPCM demodulator. shows the changes that can be made to

Not before the time division demultiplexer 90, but at the time division demultiplexer 90. DPCM demodulation is performed after the line extension rate buffers 90-1 to 90-12. Therefore, the sampling rate for performing DPCM demodulation can be lowered. Figure 7 decoding In the device, the interframe DPCM signal from the run length code unbacker 84 is It is directly applied as an input signal to the split demultiplexer 90. Y space Y, YY and Y signals LL for applying to the frequency synthesizer 91) IL' DPCM demodulation for reproducing HH is performed using interframe DPCMf This is performed by the i adjusters 101 to 104. These demodulators are full of Y information Rather than the sampling rate, it operates at half the full Y sampling rate. (Y and Y□1, are alternate L Since the scan line is interleaved with Y and YHi (H If desired, two inter-frame DPCMs correctly time division multiplexed at the scan line rate. It is possible to use a demodulator to replace the four DPCM demodulators 101 to 104. I can do it. ) Similarly, CILL 'c for applying to the CI spatial frequency synthesizer 92 , c, and CIH) I signal regeneration is full of C1 standard IHL ILH The interframe DPCM demodulators 105 to 108 operate at half the normalization speed. Therefore, it is done. C, C, C and 2 for applying to C2 spatial frequency synthesizer 93 LL 2HL 2LH C signal reproduction is at half the full C2 sampling rate 2) IH This is performed by interframe DPCM demodulators 109 to 112 operating at . one Since the C1 and 02 sampling speeds of the cup are 1/4 of the Y sampling speed of the cup, C DPCM demodulation of signals related to C1 and C2 is the standard for DPCM demodulation. The possibility of developing problems is small due to the high standardization speed.

If the high sampling speed of the inter-frame DP CMt1m device becomes a problem, the following You can take measures. The problem of reducing the speed of the DPCM demodulator for Y is In flat dimensions, using right angle mirror filtering, Y, YY and LL HL’　LH Decompose each YHH into a pair of lower and higher spatial frequency subbands, respectively. It can be solved by doing. This can be done at the transmitting station. , or output signals from speed buffers 90-1.90-2.90-3 and 90-4. It can also be done using numbers. Another delay introduced by this process is, of course, C 1 and 02 channels must be compensated. Then 8 pieces that describe Y subbands can be demodulated at 1/4 of the full sampling rate. vertical By repeating this subband method even in the dimension of The sampling rate can be further reduced. Subband method of signal describing Y to the point where the subband samples are much more sparsely sampled in the horizontal direction. interframe D of 01 and C2 samples to play C1 and C2. PCM demodulation must also proceed at a slower rate. For this reason, cl and C2 are restored. To restore Y, follow the steps described above to reduce the demodulation speed. I can do that.

The decoder in Figure 7 eliminates high spatial frequency components in areas where there are changes between frames. In Y, the procedure for stuffing zeros to make it possible is 104, demodulators 106 to 108 in C1, and demodulators 110 to 108 in C2. It also differs from the decoder of FIG. 6 in that it is performed at 112. More specifically Then, demodulators 102 to 104 are used for Y, demodulators 106 to 108 are used for C1, and Then, in C2, the predicted value is stored in each memory for each of the demodulators 110 to 112. The output signals of these demodulators are forced to zero. like this The zero-stuffing procedure generates a zero-stuffing command l by controller 86'.

J, in response to L, M, and H.

So far, in explaining the operation of the Korg in Figure 5 and the decoder in Figure 6, we have Reading from compression speed buffers 60-1 to 60-4 and line expansion speed buffer 90 Writing to -1 to 90-4 was strictly time-division multiplexed in units of scanning lines.

Linear compression speed buffers 60-5 to 6O-1ll, linear expansion speed buffers 90-5 to 60-1ll to 90-8, linear compression speed buffers 60-9 to 60-12 and linear expansion speed buffers The same procedure was used for the following.

It is preferable to implement these portions of time division multiplexing in a different manner. That's this This results in a longer run of zero values at the output of the DPCM modulator 61; This is because the compression of the coding becomes even greater. Figure 5 Time division multiplayer Instead of concatenating the compressed scan lines of the subbands of the output signal of the and the corresponding compressed scan line of each subband component of C2. One pixel sample at a time is polled sequentially, and after polling, each scan Move the line forward by one pixel. This procedure intersects the subband scan lines pixel by pixel. The number of samples in a scan line of the composite subband signal is It becomes the same as the scan line of the original video signal before being affected by the mode. composite subband Samples that are close together in the signal are the same pixels in the original video signal Or they are pixels lined up horizontally. Therefore, the zero caused by the change between frames The values of the DPCM signal that are not They tend to swarm together, rather than crowding together. In the decoder shown in Fig. 6, time division demultiplexing Writing to the linear expansion speed buffers 90-1 to 90-4 in the It is done cyclically at speed. Writing of line expansion speed buffers 90-5 to 9G-8 The same is true for writing to the linear expansion speed buffers 90-9 to 90-12. Ru.

When dividing Y into subbands, the subbands have the same density for C1 and C, or horizontally sampled with the same density as subbands C1 and C2 and non-zero The idea of arranging samples so that the DPCM samples are clustered has been pushed further. Rukoto can. Further refinement in the time division multiplexer of the Korg and decoder A built-in speed buffer circuit allows Y, CI and C2 samples to describe the same part of the image. the digital sample slider fed to the interframe DPCM modulator. They can be placed close to each other in the room. One block of image data (i.e., time All subband signals in one cycle of split multiplexer operation are compressed. The number of samples per scan line is the same, and by polling it sequentially, Digital data stream fed to Korg's differential pulse code modulator It is possible to generate a problem. In the decoder, after the differential pulse code demodulator, A time-division decomposition operation that complements the time-division multiplexing in the multiplexing circuit ensues.

Figure 8 shows the spatial frequency analyzer at the transmitting station and the spatial frequency synthesizer at the receiving station. shows a cog that is included in the predictor part of the DPCM modulator at the transmitting station. . The digitized Y signal from the analog-to-digital converter 41 is sent to the subtracter 1. 41 minuend input port from which the previous frame's Y predicted value is subtracted. be done. This predictor is applied to the subtraction input port of subtractor 141. Subtractor 14 The Y inter-frame error signal appearing on the difference output port of 1 is, as far as the delay length is concerned. , the operation of compensation delay device 144 similar to compensation delay device 44 in the Korg of FIG. A delayed error signal is supplied to the Y spatial frequency synthesizer 145.

The spatial frequency analyzer 145 processes the inter-frame error of Y, not Y itself. This is different from the Korg spatial frequency analyzer shown in FIG. Similarly, spatial frequency analyzer 1 46 and 147 are C1 supplied from the difference output ports of subtracters 142 and 143, respectively. and C2 interframe errors are processed.

The time division multiplexer 160 is the same as the Korg time division multiplexer 60 in FIG. However, the point that acts on the interframe error of Y + CI and C2 is different. The time division multiplexed output signal from the multiplexer 160 is sent to the quantizer 161. the response is the differential pulse code modulator output signal and the run length code is is supplied to the code assembler 62. Elements B2 to 65 are shown in FIG. It operates similarly to the elements with certain corresponding reference numerals. code output speed buffer The built-in controller 166 has all the functions of the v4 controller 55 in the Korg in the 1#I5 diagram, It has some functions of the controller 86 in FIG. 6, which will be explained later. do.

The Korg code output rate buffer 65 of FIG. 8 is the same as a portion of the decoder of FIG. It sends a code output for transmission to various decoders. Controllers 166.86 are correctly cooperating. Assuming that the Korg elements 62-65 of FIG. 8 or the decoder of FIG. No amplitude error due to quantization enters the elements 81 to 84, so the transmission error Otherwise, the output signal from run length code unbacker 4 of FIG. It should be the same as the output signal from 161. Therefore, the differential pulse in Figure 8 From the feedback connection forming the predicted value portion of the code modulator, elements 62-65 and The children 81 to 84 are used as models, and the child elements are omitted. The output signal of the quantizer 161 is device 112167. This device performs DPCMffl modulation of the signal. Then, the demodulated signal is decomposed to separate the subband spectra, Y, CI and C2 Subband spectrum is fed to a spatial frequency synthesizer and subtracted to recover the signal Vessel 141. ! In order to make it a prediction signal to be applied to the subtraction input port of 42,143 , delaying the reproduced Y, CI and C2 signals.

FIG. 9 shows a configuration of the device 167 of FIG. 8 which is similar to certain parts of the decoder of FIG. It is shown in some detail. Interframe DPC similar to demodulator 85 of the decoder of FIG. The M demodulator 185 responds to the interframe DP CM signal from the quantizer 151. to recover the interleaved subband samples of Y, C, and C2. . A time division demultiplexer similar to the time division demultiplexer 90 of the decoder of FIG. Y spatial frequency synthesizer 191 then reproduces the delayed Y signal using 01, then the spatial frequency synthesizer 192 describes the delay C 1 subband describing C1, and then C2 spatial frequency to reproduce the signal. The number synthesizer 193 separates subbands describing C2 to reproduce the delayed C2 signal. Let go. The combiners 191, 192, 193 are the combiners 91, 92, . . . 92 of the decoder in FIG. They have the same structure as 93. The digital delay line 194 is a Y spatial frequency synthesizer. Analog-to-digital conversion is performed by increasing the delay of the Y signal restored by 191. The time appears to be one frame later than the Y signal supplied from the subtracter 141 from the subtracter 41. Make it. Digital delay line 195 is restored by C1 spatial frequency synthesizer 192 The subtractor increases the delay of the C1 signal obtained by analog to digital converter 42. The time is delayed by one frame from the C1 signal supplied to 142.

C2 where the digital delay line 196 is restored by the 02 spatial frequency synthesizer 193 Increases the delay of the signal and supplies it from the analog-to-digital converter 43 to the subtracter 143. The time is set to be one frame later than the supplied C2 signal.

The restored Y, CI and C2 signals are input to the subtraction input buttons of subtracters 141, 142 and 143. The Korg differential pulse code modulation loop shown in FIG. closes.

Write control signals A to H for speed buffers 190-1 to 190-12 and The read control signals T through W are generated by fill controller 166. This is the amount Starting from the output signal of the daughterizer 161, the decoder controller 86 of FIG. You can proceed in the same way as before.

A fill controller 166 controls the Y, CI and C2 signals with the code output from the Korg of FIG. It also controls the zero-filling procedure when the signal does not represent high spatial frequency components. space When omitting high frequency components, the padding controller 166 controls the The output signal of the quantizer 161 is suppressed. All output samples of the high spatial frequency component scan line will have a value of zero. .

A padding controller 166 controls the recovery used to close the interframe DPCM modulator loop. Connection 186 to interframe DPCM demodulator 185 in encoder unit 167 A control signal is sent via the interframe DPCMf controller 85, and this control signal is Instruct the packing procedure. At this time, the output signal and the prediction that should be delayed by one frame. Both signal and signal are forced to zero value.

By replacing the decoder device 167 with a decoder device of the type described below. , the cog in FIG. 8 can be modified. The output signal of quantization 5161 is the seventh Inter-frame DP distributed to CMt131 units as shown in 101 to 112 in the figure Therefore, the signal is supplied to a time division multiplexer as shown at 90 in FIG. These demodulations The subbands from the receiver are routed to a spatial frequency synthesizer as shown at 91 to 93 in Figure 7. divided.

The Y, CI and C2 signals reproduced by the spatial frequency synthesizer are the original Y, CI and C2 signals, and are delayed by 17 frames in time from the subtracter 1. 41,142°143 is applied to the subtraction input port. Analog to DPCM loop It is also possible to modify the Korg in Figure 8 to include digital converters 41, 42, 43. It is Noh.

Using a slightly more specific form of zero stuffing method than what has been explained so far, you can calculates the amount of high spatial frequency detail that is lost in parts of an image that do not change from frame to frame. Embodiments of the invention are also possible with reduced numbers. In this alternative method, the Y error signal Low spatial frequency components (i.e. Y DPCML The signal) is detected by a threshold value in both the encoder at the transmitting station and the decoder at the receiving station, Detecting larger values of the Y error signal that describe areas with pixel changes between frames Ru. In these areas, pixel samples of high spatial frequency components from the previous frame The predicted value may be wrong, and the differential pulse φ code modulator and the differential pulse The memory of the code demodulator is instead padded with zeros as the predicted value. flame In areas where no change in time is detected, no zero filling is performed to replace the sample. Therefore, details with high spatial frequencies are preserved.

This alternative zero-stuffing procedure uses low-sky areas where motion occurs in a series of image frames. The fact that high spatial frequencies are not resolved temporally or spatially as much as the intermediate frequencies Make it possible to utilize.

An embodiment of the invention is also possible in which only the luminance component of visible light is digitally transmitted. mosquito The camera uses infrared spectral energy instead of or in addition to visible light energy. Embodiments of the invention that are sensitive to radiation or ultraviolet energy are also possible.

An embodiment of the invention uses a lookahead method to influence process steps. It is possible. In this invention, the coding procedure adopted is symbolic coding. Examples are also possible. This is a run length coding procedure that generates a modified DPCM. is carried out outside the

+90 Submission of Translation of Written Amendment (Article 184-8 of the Patent Law) - September 2018 Above, indication of patent application PCT/US 88104241 , name of invention Interframe DP of time-division multiplexing spatial frequency analysis of video signals Proceeding from CM Occurrence of symbol code 3, patent applicant Address: Schenectaday, New York, 12345, United States River Road, No. 1 Name General Electric Company Representative King, AR Sir M Nationality: United States of America 4. Agent Address: 107 4th floor, No. 35 Kowa Building, 1-14-14 Akasaka, Minato-ku, Tokyo Japan General Electric Co., Ltd., Far East Patent Department, October 1989 25th day 6. List of attached documents (1) Translation of written amendment 1 request The range of 1. Perform interframe differential pulse code modulation (DPCM) of component video signals means for performing run length encoding of samples having a value of zero from the DPCM modulated signal; and a non-zero value of the run length code and modulation and run length encoded signal. It has a means to perform statistical coding of samples and multiple component videos. Digitally televised - successive image frames described by signals Differential pulse code modulation, run length encoding and statistical code prior to the coding step, having means for time division multiplexing the component video signals together. transmitting device.

2. Claim 1, wherein the time division multiplexing means includes means for multiplexing each scanning line. transmitting device.

3. Time division multiplexing of different component videos that are at least spatially substantially corresponding; Claims made by interleaving the scan lines of the signal pixel by pixel 1. The transmitting device according to 1.

4. furthermore - at least one digital video describing successive image frames; In response to each signal, we perform spatial frequency analysis of each digital signal. For a video signal, at least some subband signals in each set generate it. so that it is generated at a lower sampling rate than the digital video signal used to means for generating one set of subband signals, and a means for generating a set of subband signals from each subband signal. A differential pulse code modulation (DPCM) signal that generates a differential pulse code modulation (DPCM) signal in interframe format. a pulse code modulator and a differential pulse code modulated signal in interframe format; Interleave scan lines in time to form a digital sample stream a time-division multiplexer and a time division multiplexer to and a symbol code that generates a symbol code describing the pull. Transmitting device.

5. The means for performing the spatial frequency analysis is configured to analyze luminance fluctuations in the subsequent image frames. A relatively wideband digital video signal that describes Perform spatial frequency analysis on two digital video signals to obtain the above-mentioned results. 5. A transmitting device according to claim 4, which completes the description of all colors of an image frame.

6. said symbol Korg has a value of at least zero in said sample stream; means for encoding a run of samples with a run length to generate a modified DPCM signal; A looker that encodes the modified DPCM signal in a unified manner and generates the symbol code. 5. The transmitting device according to claim 4, further comprising a top table.

7. A method for broadcasting subsequent image frames using the transmitting device according to claim 4. at least one digital video signal relying on a television camera source. and transmitting each of the at least one video signal as claimed in claim 4. and transmitting said symbol code via a broadcast transmission medium. Law.

8. Digitally televised - symbol code describing successive image frames A receiving device for the signal converts the symbol code into a differential pulse code in interframe format. a symbol decoder for generating a sample stream of a DPCM signal; The digital sample stream is recorded as successive image frames in the preceding sequence. a differential spatial frequency analysis of each of the at least one digital video signal; When dispersing into each subband signal, which is a pulse code modulated component, a division demultiplexer provided for each subband signal; a respective interframe DPCM decoder for restoring spatial frequency components; In response to the respective spatial frequency analysis of each of the one digital video signal, and means for synthesizing digital video signals.

9. The symbol decoder converts the symbol code into a non-zero value in the DPCM signal. a stream of samples with a value of zero in the DPCM signal and a sample with a value of zero in the DPCM signal. and a lookup table to convert the run length code of the run to the run length code of the run. In response to run samples with zero values between samples with non-zero values 9. The receiving device according to claim 8, further comprising means for restoring the DPCM signal by distributing the DPCM signal into Device.

10, after the means for performing interframe differential pulse code modulation of the component video signals; , followed by a means for run-length encoding of samples with values of zero, followed by run-length encoding of samples with values of zero. A means of performing statistical expression coding of samples with code and non-zero values follows. a format in which a digital text is described by multiple component video signals. Revised - In the transmitter for subsequent image frames, the statistical coder The differential pulse-code modulation responses of the component video signals are time-division multiplexed before the coding procedure. a transmitting device having means for converting

11. The transmitting device according to claim 10, wherein the time division multiplexing is performed for each scanning line. .

12. The means for time division multiplexing is at least substantially spatially corresponding to different Claim 1 comprising means for interleaving scan lines of the component video signals pixel by pixel. Transmitting device described in 0.

13, furthermore: - at least one digital video describing successive image frames; In response to each digital signal, spatial frequency analysis is performed on each digital signal. For a video signal, at least some subband signals in each set are generated at a lower sampling rate than the digital video signal used to generate the means for generating each set of subband signals in such a manner that the subband signals temporally interleaving the scanlines of the digital sample stream. from the digital sample stream; A differential pulse code modulation (DPCM) signal that generates a differential pulse code modulation (DPCM) signal in interframe format. A pulse code modulator and the samples in the digital sample stream. and a symbol hoop for generating a symbol code to describe. Device.

14. The means for performing spatial frequency analysis is responsive to said - continuation image length code. DPC is performed by dispersing the runs of samples with values among the samples with non-zero values. 22. The receiving device according to claim 5, further comprising means for restoring the M signal.

23, the at least one digital video signal is a relatively wideband luminance video signal; and two relatively narrowband chrominance video signals. Receiving device according to range 21.

24. Digitally televised - transmitter for subsequent image frames has a number (at least one) of subtractors, each subtractor having - a subsequent image file; Each minuend input baud receives a respective digital video signal describing the frame. for each digital video signal received on each minuend input port. Respective subtraction input ports that receive prediction signals and provide respective digital error signals. and - each data frame describing the subsequent image frame. In response to the digital error signal, we perform a spatial frequency analysis of the signal in each set. At least some of the subband signals are connected to the digital signal used to generate them. one set of subbands, each generated at a lower sampling rate than the video signal. means for generating a subband signal for each digital error signal; , time-division multiplexed as a digital sample stream from means for generating a pulse code modulated response and the digital sample stream; In response to the request for confirmation of the international search report a quantizer that generates a digital sample stream that is coarsely quantized; As the output signal of the transmitting device, the coarser quantized digital sample A symbolic code that generates a symbolic code that describes the stream, and then further coarsely quantized. the respective subtract inputs of each subtractor from the digital sample stream and a device for generating respective predicted values for the ports.

25. Generate a respective preset signal for each subtractor capo of each subtractor. a further coarsely quantized digital sample stream from said quantizer; demodulates each subband into a set of subbands, each describing a digital video signal. Interframe differential pulse code demodulation to generate time-division multiplexed signals including a spatial frequency synthesizer for one set of subband signals, and a spatial frequency synthesizer for each set of subband signals; From the supplied time division multiplexed signal, it is applied to each spatial frequency synthesizer and a set of subband signals each generating a digital video signal synthesized from and a time-division demultiplexer that separates each digital video signal. and means for delaying the respective prediction signals. Transmitting device as described.

Claims (25)

[Claims] 1. has a value of zero following interframe differential pulse code modulation of the component video signal. Run length encoding of one sample followed by run length code and zero. This is a format that performs statistical formula coding for samples with different values, and a series of digitally televised images described by component video signals In a transmitting device for frames, the component video signals are converted into differential pulse codes. Before each step of modulation, run length encoding and statistical coding, time division multiplexing is performed together. Duplicated transmitter. 2. 2. The transmitting device according to claim 1, wherein said time division multiplexing is performed for each scanning line. 3. Time division multiplexing provides at least spatially substantially corresponding distinct component videos. Claims made by interleaving the scan lines of the signal pixel by pixel 1. The transmitting device according to 1. 4. Additionally, at least one digital video describing a series of image frames. In response to each signal, spatial frequency analysis is performed on each digital signal. For a video signal, at least some subband signals in each set generate it. so that it is generated at a lower sampling rate than the digital video signal used to means for generating one set of subband signals, and a means for generating a set of subband signals from each subband signal. A differential pulse generator that generates a differential pulse code modulation (DPCM) signal in interframe format. a pulse code modulator and scanning of the differential pulse code modulated signal in the interframe format; Interleave lines in time to form a digital sample stream A time division multiplexer and the samples in the digital sample stream and a symbol coder that generates a symbol code describing the transmission according to claim 1. Device. 5. The means for performing the spatial frequency analysis analyzes luminance fluctuations in the series of image frames. A relatively wideband digital video signal that describes Perform spatial frequency analysis on two digital video signals to 5. A transmitting device according to claim 4, which completes the description of all colors of an image frame. 6. the symbol coder has a value of at least zero in the sample stream; means for encoding a run of samples with a run length to generate a modified DPCM signal; A lookup that encodes the modified DPCM signal into a statistical formula to generate the symbol code. 5. The transmitting device according to claim 4, further comprising a pull table. 7. A method for broadcasting a series of image frames using the transmitting device according to claim 4. at least one digital video signal relying on a television camera source. and transmitting each of the at least one video signal as claimed in claim 4. and transmitting said symbol code via a broadcast transmission medium. Law. 8. a symbolic code that describes a series of digitally televised image frames A receiving device for the system converts the symbol code into a differential pulse code in interframe format. a symbol decoder for generating a code modulated (DPCM) signal; at least one stream describing successive image frames in said sequence; Differential pulse code variation of each spatial frequency analysis of one digital video signal Time-division demultiplexing that separates each subband signal, which is a modulated component, into each subband signal. It is provided for each subband signal, and the spatial frequency analysis component is a respective interframe DPCM decoder for restoring the at least one digital the digital video in response to respective spatial frequency analysis of each of the video signals; and a means for combining signals. 9. The symbol decoder converts the symbol code into a non-zero value in the DPCM signal. a stream of samples with a value of zero in the DPCM signal and a sample with a value of zero in the DPCM signal. and a lookup table to convert the run length code of the run to the run length code of the run. In response to run samples with zero values between samples with non-zero values 9. The receiving device according to claim 8, further comprising means for restoring the DPCM signal by distributing the DPCM signal into Device. 10. has a value of zero after interframe differential pulse code modulation of the component video signals. Run length encoding of one sample, followed by run length code and non-zero value. A format that performs statistical coding of samples with multiple component videos. A series of digitally televised image frames described by a signal The response of differential pulse code modulation to component video signals in a transmitting device for However, before the run length encoding and statistical formula coding steps, time division multiplexing is performed for each scan line. transmitting device. 11. 11. The transmitting device according to claim 10, wherein the time division multiplexing is performed for each scanning line. . 12. said time-division multiplexing comprises at least spatially substantially corresponding distinct component videos; Claims made by interleaving the scan lines of the signal pixel by pixel 10. The transmitting device according to 10. 13. Additionally, at least one digital video recording sequence describes a series of image frames. In response to each digital signal, spatial frequency analysis is performed on each digital signal. For a video signal, at least some subband signals in each set are generated at a lower sampling rate than the digital video signal used to generate the means for generating each set of subband signals in such a manner that the subband signals to create a digital sample stream by temporally interleaving the scan lines of a time division multiplexer to form a digital sample stream; Differential pulses that generate differential pulse code modulation (DPCM) signals in frame-to-frame format Describes the code modulator and the samples in the digital sample stream. 11. The transmitting device according to claim 10, further comprising a symbol coder that generates a symbol code. 14. Means for performing spatial frequency analysis calculates luminance variations in the series of image frames. Describe relatively wideband digital video signals and analysis is performed on two digital video signals to determine the sequence of image frames. 14. The transmitting device according to claim 13, which completes the description of all colors of the system. 15. The symbol coder detects at least a zero value in the sample stream. means for encoding a run of samples with a run length to generate a modified DPCM signal; A lookup that encodes the modified DPCM signal into a statistical formula to generate a symbol code. - The transmitting device according to claim 13, comprising: a table. 16. The transmitting device according to claim 13 and the transmitting device via a digital transmission line. a series of image frames comprising at least one receiving device coupled to a transmitting device; In the system for digitally televising a broadcast program, the at least one receiving device is a differential pulse code modulation (DPCM) in interframe format from the symbol code. a symbol decoder that generates a signal from the DPCM signal, and a symbol decoder that generates a subband in the series; A digital sample stream consisting of interleaved scanlines of a signal a DPCM decoder that generates a digital sample stream; at least one image frame separated into sub-band signals to describe successive image frames in said sequence. A time-sharing hoax that performs spatial frequency analysis of each of a single digital video signal. a respective one of the multiplexers and each of the at least one digital video signal; means for synthesizing the digital video signal in response to spatial frequency analysis; method. 17. Broadcasting a series of image frames using the transmitting device according to claim 13 The method relies on a television camera source to provide at least one digital generating video signals, each of the at least one video signal as defined in claim 13; each of which is applied to the described transmitting device and transmits said symbol code via a broadcast transmission medium. A method involving stages. 18. A symbolic code that describes a series of digitally televised image frames. The receiving device for the code converts the symbol code into a differential pulse code in interframe format. a symbol complexer that generates a code modulated (DPCM) signal; A digital sample consisting of a series of interleaved subband signals. an interframe DPCM decoder that generates a digital sample stream; Separating the pull stream into respective sub-band signals to separate successive images in the series. a spatial frequency of each of the at least one digital video signal describing a frame; a time division demultiplexer for performing numerical analysis; and the at least one digital demultiplexer; the digital video signal in response to respective spatial frequency analysis of each of the video signals; and means for synthesizing the signals. 19. The symbol decoder converts symbol codes into non-zero values in the DPCM signal. a stream of samples with a value of zero in the DPCM signal and a stream of samples with a value of zero in said DPCM signal. a lookup table to convert the run length code into the run length code; In response, run samples with zero values between samples with non-zero values. 19. The receiver according to claim 18, further comprising means for restoring the DPCM signal in a distributed manner. communication device. 20. The at least one digital video signal is a relatively wideband luminance video signal. The video signal consists of an audio signal and two relatively narrowband chrominance video signals. Receiving device according to claim 18. 21. A symbolic code that describes a series of digitally televised image frames. A receiving device for the symbol code converts the composite differential pulse in interframe format from the symbol code. A symbol decoder that generates a code modulated (DPCM) signal and a multiplexer of the interframe format. at least one combined DPCM signal describing successive image frames in said sequence; Describe each subband signal for spatial frequency analysis of a digital video signal. a time division demultiplexer that separates the component DPCM signals into interframe format components; generating respective subband signals from each interframe format component DPCM signal; a respective DPCM multiplexer and a plurality of said digital video signals in response to respective spatial frequency signals; and means for synthesizing video signals. 22. The symbol decoder converts symbol codes into non-zero values in the DPCM signal. a stream of samples with a value of zero in the DPCM signal and a stream of samples with a value of zero in said DPCM signal. a lookup table to convert the run length code into the run length code; In response, the runs of samples with zero values are separated between the samples with non-zero values. 22. The receiving apparatus according to claim 21, further comprising means for restoring the DPCM signal by dispersing the DPCM signal. 23. The at least one digital video signal is a relatively wideband luminance video signal. and two relatively narrowband chrominance video signals. Receiving device according to range 21. 24. Transmitter for a series of digitally televised image frames In this case, there are a number of subtractors (at least one), each subtractor having a sequence of image frames. a respective minuend input port that receives a respective digital video signal describing the program. , for each digital video signal received at each minuend input port. a respective subtraction input port for receiving a measurement signal and a respective subtraction input port for providing a respective digital error signal; each has a difference output port that describes a series of image frames. In response to the digital error signal, spatial frequency analysis is performed to determine each digital error signal. For the digital error signal, at least some subband signals in each set are generated at a lower sampling rate than the digital video signal used to generate the means for generating each set of subband signals in such a manner that the subband signals to them, time-division multiplexed as a digital sample stream. means for generating a digital pulse code modulation response for the digital pulse code modulation; Coarsely quantized digital samples in response to a sample stream ・A quantizer that generates a stream and a coarser quantizer as the output signal of the transmitter A notation that generates the symbolic code that describes the encoded digital sample stream. From the encoder/decoder and a coarser quantized digital sample stream, each and a device for generating respective predicted values for respective subtraction input ports of each subtractor. transmitting device. 25. A device for generating respective prediction signals for respective subtraction input ports of each subtractor. is the coarser quantized digital sample stream from the quantizer. demodulate the video signal into a set of subbands, each describing a digital video signal. An interframe differential pulse code demodulator that generates a time division multiplexed signal containing a code signal. and a spatial frequency synthesizer for one set of subband signals, respectively, and a spatial frequency synthesizer provided from the demodulator. From the supplied time division multiplexed signal, it is applied to each spatial frequency synthesizer and each set of subband signals to generate a digital video signal synthesized from Separating time division demultiplexer and each combined digital video signal and means for generating the respective prediction signals by delaying the prediction signals. Transmitter on board.